mcpField
D
debugging-dags

debugging-dags

github.com/necatiarslanData & databases4446 installsUpdated 21w ago
Repository

About debugging-dags

A skill for data engineers investigating failed Airflow DAG runs. It guides diagnosis from failure discovery through root-cause analysis, impact assessment, remediation, and rerun guidance using extension tools.

Key features

  1. 01Find recent failed DAG runs
  2. 02Inspect task errors and logs
  3. 03Compare runs and review DAG source
  4. 04Assess impact and suggest fixes
  5. 05Provide prevention and rerun guidance

Use cases

  1. 01Investigate a named DAG failure
  2. 02Locate failures across DAGs
  3. 03Prepare remediation and rerun steps

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/necatiarslan/airflow-vscode-extension --skill debugging-dags

After installing, type/debugging-dagsin Claude Code to confirm it loaded.

Details

From the project README

DAG Diagnosis

You are a data engineer debugging a failed Airflow DAG. Use the extension tools to identify root cause and provide actionable remediation.

Step 1: Identify the Failure

If a specific DAG was mentioned:

  • Use get_dag_runs to find recent failed runs
  • If the latest failed run is sufficient, use analyse_dag_latest_run

If no DAG was specified:

  • Use get_failed_runs to list recent failures across DAGs
  • Ask which DAG to investigate further

Step 2: Get Error Details

Once a failed run is identified:

  1. Use analyse_dag_latest_run or get_dag_run_detail
  2. Focus on the failed task logs in the analysis
  3. Categorize the failure:
    • Data issue
    • Code issue
    • Infrastructure issue
    • Dependency issue

Step 3: Check Context

Gather context to understand why this happened:

  • Compare with prior runs using get_dag_runs or get_dag_history
  • Review DAG code via get_dag_source_code
  • Check current system status using go_to_server_health_view

Step 4: Provide Actionable Output

Structure your diagnosis as:

Root Cause

Be specific about what failed and why.

Impact Assessment
  • Which tasks or outputs are affected
  • Whether downstream consumers are blocked
Immediate Fix

Concrete steps or code changes.

Prevention

Data checks, retries, alerting, or code hardening.

Rerun Guidance
  • Trigger a rerun using trigger_dag_run

Notes

  • Use go_to_dag_log_view when a deep log inspection is needed.
  • Avoid CLI commands for Airflow inspection.

Comments

Markdown supported